donor_classifier

practice deploying a ML model

Goal: Assuming that a charity realizes that people who make more than $50k/yr are more likely to donate and this charity organization wishes to identify potential donors, predict whether income exceeds $50K/yr based on census data.

Data source: https://archive.ics.uci.edu/ml/datasets/Adult

1. Explorartory data analysis

• documented here

2. ML model building

• traditional ML models documented here
• neural net model documented here

3. ML model as a RESTful API

• actual server app
• the API is deployed on Heroku (https://donorclf.herokuapp.com/predict)

Sending a request to the API:

example:

We can send a request to https://donorclf.herokuapp.com/predict and request for a prediction for the following donor:

feature	value
age	39
workclass	State-gov
fnlwgt	77516
education	Bachelors
education_num	13
marital_stat	Never-married
occupation	Adm-clerical
relationship	Not-in-family
race	White
sex	Male
capital_gain	2174
capital_loss	0
hr_per_wk	40
native_country	United-States

Note: information on the features can be found here

The request should be sent in json format

{
    "age": 39,
    "workclass": " State-gov",
    "fnlwgt": 77516,
    "education": " Bachelors",
    "education_num": 13,
    "marital_stat": " Never-married",
    "occupation": " Adm-clerical",
    "relationship": " Not-in-family",
    "race": " White",
    "sex": " Male",
    "capital_gain": 2174,
    "capital_loss": 0,
    "hr_per_wk": 40,
    "native_country": " United-States"
}

The API should return a prediction in json format

{
    "prediction": [
        0
    ]
}

example2:

request

{
    "age": 34,
    "workclass": " Private",
    "fnlwgt": 77516,
    "education": " Masters",
    "education_num": 20,
    "marital_stat": " Married-civ-spouse",
    "occupation": " Tech-support",
    "relationship": " Husband",
    "race": " Asian-Pac-Islander",
    "sex": " Male",
    "capital_gain": 2000,
    "capital_loss": 0,
    "hr_per_wk": 80,
    "native_country": " Canada"
}

returns a prediction

{
    "prediction": [
        1
    ]
}

If the prediction is 0, then the model classifies the donor as a donor that makes LESS than $50k annually.

If the prediction is 1, then the model classifies the donor as a donor that makes MORE than $50k annually.

Next step:

1. I left some sanity check code chunks in the EDA and modelling parts. I haven't been able to rewrite them as proper unit tests yet.
2. The API needs a more through testing as well.
3. Input authentication has not be implemented yet.